This article was downloaded by: [Rebecca Treiman]On: 08 July 2011, At: 07:53Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

Scientific Studies of ReadingPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/hssr20

Are Young ChildrenLogographic Readers andSpellers?Margo Bowman a & Rebecca Treiman ba Wayne State Universityb Washington University in St. Louis

Available online: 16 May 2008

To cite this article: Margo Bowman & Rebecca Treiman (2008): Are Young ChildrenLogographic Readers and Spellers?, Scientific Studies of Reading, 12:2, 153-170

To link to this article: http://dx.doi.org/10.1080/10888430801917233

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching and private study purposes.Any substantial or systematic reproduction, re-distribution, re-selling, loan,sub-licensing, systematic supply or distribution in any form to anyone isexpressly forbidden.

The publisher does not give any warranty express or implied or make anyrepresentation that the contents will be complete or accurate or up todate. The accuracy of any instructions, formulae and drug doses should beindependently verified with primary sources. The publisher shall not be liablefor any loss, actions, claims, proceedings, demand or costs or damages

whatsoever or howsoever caused arising directly or indirectly in connectionwith or arising out of the use of this material.

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

Are Young Children LogographicReaders and Spellers?

Margo BowmanWayne State University

Rebecca TreimanWashington University in St. Louis

According to many views of literacy development, prereaders use a logographic ap-proach when they attempt to link print and speech. If so, these children should findpairs in which the spelling–pronunciation links are consistent with their writing sys-tem no easier to learn than arbitrary pairs. We tested this idea by comparing the abil-ity of U.S. prereaders (M age = 4 years 9 months) to learn phonetically motivatedpairs like AP–ape and MA–may and arbitrary pairs like OM–ape and PO–may. Inboth spelling and reading tasks, children learned the pairs with vowel letter namecues more easily than the arbitrary pairs. Phonetically motivated pairs were espe-cially advantaged when the vowel letter names were at the beginning (e.g., AP–ape)rather than the end (MA–may). Prereaders who have some knowledge about letters,as U.S. preschoolers typically do, are not limited to a logographic approach in learn-ing about print.

The phonological forms of spoken words bear an arbitrary relationship to theirmeanings. For example, there is no reason why a chair should be called /t εr∫ / asopposed to something else. Indeed, a chair is called a / |sija/ in Spanish. (For an ex-planation of the phonetic symbols used here, see International Phonetic Associa-tion, 1999.) Learners of spoken languages must memorize the arbitrary phonologi-cal forms that are associated with the concepts of chair, dog, and so on. There areno meaningful or systematic links that they can use to motivate the associations be-tween the phonological and semantic forms. The situation is different for learners

SCIENTIFIC STUDIES OF READING, 12(2), 153–170Copyright © 2008 Taylor & Francis Group, LLCISSN: 1088-8438 print / 1532-799X onlineDOI: 10.1080/10888430801917233

Correspondence concerning this article should be addressed to Rebecca Treiman, Psychology De-partment, Washington University, St. Louis MO 63130. E-mail: rtreiman@wustl.edu

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

of most written languages. In alphabetic and syllabic writing systems, writtenwords are linked in systematic ways to their pronunciations. For example, thespelling chair is related to the phonological form /t εr∫ / by virtue of spelling–sound correspondences that hold across many English words. The spelling is moti-vated rather than arbitrary for those who know that ch corresponds to /t /∫ in wordssuch as chip and that air corresponds to /εr/ in words such as pair. Becoming agood reader involves learning about the spelling–sound mappings of the language.Skilled readers reveal their knowledge of these links when they pronounce a previ-ously unseen word like glair on first exposure or when they hesitate on a word likechoir, which deviates from the typical spelling-to-sound mappings of the language.

We know that skilled readers benefit from the systematic links between lettersand pronunciations that exist in alphabetic writing systems, but we know less abouthow and when children begin to do this. According to several influential views ofliteracy development (e.g., Byrne, 1992; Ehri, 1998; Frith, 1985), children typi-cally learn their first words in a logographic or prealphabetic way. Rather than us-ing systematic links between letters and sounds, children identify each printedword on the basis of its salient visual characteristics. For example, a child may rec-ognize dog by virtue of the “tail” at one end of the printed word. This approach toword identification is quite limited. Logographic learners often confuse printedwords with one another, and they tend to forget them quickly. According to theseinfluential theories of literacy development, literacy learning begins to surge onlywhen children start to take advantage of alphabetic links between printed and spo-ken words.

Ehri and Wilce (1985) found evidence for a prealphabetic approach among chil-dren who they classified as prereaders when they taught U.S. 5-year-olds to readtwo sets of printed items. The items in one set contained letters that correspondedto the sounds in the words’ pronunciations. In this condition, for example, MSKwas presented as a spelling for mask. The items in the other set violated conven-tional letter-sound correspondences but were more visually distinctive. In this con-dition, for example, UHE was taught as giraffe. Children had up to 10 trials to learnthe pronunciations of the items in each set. Children who could read no more thanone word from a list of simple words such as look, it, and stop performed better onthe arbitrary but visually distinctive spellings than on the phonetically motivatedspellings. According to the researchers, these children were relying on a logo-graphic approach to link print and speech. Abreu and Cardoso-Martins (1998)found similar results among Portuguese-speaking prereaders who had little knowl-edge about letters, although children with more letter knowledge performed differ-ently (see also Cardoso-Martins, Resende, & Rodrigues, 2002). In the study byEhri and Wilce, as in studies by Rack, Hulme, Snowling, and Wightman (1994),and Laing and Hulme (1999), children who could read at least a few words aloudlearned phonetically motivated items more easily than arbitrary or less well moti-vated items.

154 BOWMAN AND TREIMAN

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

The results of Ehri and Wilce (1985) suggest that young children start out by us-ing a logographic approach to learn about print. A potential problem with this con-clusion, however, stems from the fact that the nature of the spelling–sound linkswas confounded with the visual characteristics of the print in Ehri and Wilce’sstudy. The phonetically motivated spellings looked less distinctive than the arbi-trary ones. Thus, it is not clear whether prereaders’ failure to perform better on thephonetically motivated spellings than the arbitrary ones reflects a failure to use al-phabetic mappings or a preference for spellings that are visually distinctive. Ifprereaders approach print logographically, they should perform no better on alpha-betically motivated spellings than on arbitrary spellings when the two types ofspellings are similar in appearance. We tested this prediction in our study.

A further limitation of the studies described so far is that they examined learn-ing in the spelling-to-sound direction, or reading, but not learning in the sound-to-spelling direction, or spelling. A complete understanding of literacy develop-ment, of course, requires consideration of both reading and spelling. Frith (1985)proposed that children move from a logographic approach to an alphabetic ap-proach for spelling before they do so for reading. If so, we would expect to find dif-ferences in children’s pattern of performance in spelling and reading tasks. Twostudies (Bowman & Treiman, 2002, Experiment 2; Treiman, Sotak, & Bowman,2001, Experiment 3) have examined learning in the sound-to-spelling direction us-ing procedures similar to those just described. However, direct comparisons be-tween spelling and reading have been hindered because the letters used in theseprevious spelling experiments differed in their visual characteristics from the let-ters used in the reading experiments. For example, the earlier experiments studiedlearning in the sound-to-spelling direction for visually distinctive but alphabeti-cally arbitrary spellings by using items in which the letters differed from one an-other in color as well as other attributes. In the experiments studying learning in thespelling-to-sound direction, however, the letters in these items differed in size andpositioning but not in color. Our study examined children’s learning of phoneti-cally motivated and arbitrary print–sound associations in both spelling and read-ing, using stimuli that were well matched across the two tasks.

Our study was based on that of Bowman and Treiman (2002, Experiment 3),which as far as we know is the only study to have examined how children who can-not read simple words learn alphabetically motivated and arbitrary print–soundpairs when the two types of stimuli are equated for their visual characteristics. Thatstudy examined learning in the spelling-to-sound direction only. In the phoneti-cally motivated conditions, children learned that a letter string such as LK was pro-nounced as elk, a word with a VCC (V = vowel, C = consonant) structure. Childrenalso learned that a letter string such as FL was pronounced as fell, a word with aCVC structure. In the VCC and CVC phonetically motivated conditions, both con-sonants of the spoken words were spelled with plausible letters. The entire name ofthe first printed letter was heard in the spoken VCC words and the entire name of

LOGOGRAPHIC READING AND SPELLING 155

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

the last printed letter was heard in the CVC words. When elk is spelled as LK, forexample, L can be interpreted as representing both phonemes in its name—/ε/ and/l/. Letter names were included in the stimuli because U.S. children learn thenames of the letters from an early age (e.g., Worden & Boettcher, 1990) and be-cause they appear to use this knowledge in linking print and speech (e.g., Foulin,2005; Treiman & Rodriguez, 1999). The arbitrary conditions of the Bowman andTreiman study included pairs such as LT–ark and ML–far. No letter name cluesthat could aid performance appeared in these items; the letters in the printed itemswere not plausible representations of the sounds in the spoken items.

Bowman and Treiman (2002, Experiment 3) tested children who had a meanage of 4 years 7 months and who could not read any simple words on a screeningtest. Overall, the children performed better in the phonetically motivated condi-tions than the arbitrary conditions. For the VCC motivated condition, where thename of the first printed consonant letter was heard in the corresponding spokenword (e.g., LK–elk), the superiority for the motivated condition over the arbitrarycondition was statistically significant. This superiority was smaller and not statisti-cally reliable for the CVC items, where the name of the second consonant letterwas heard (e.g., FL–fell). The children in the Bowman and Treiman study, likemost U.S. children of this age, knew the names of a number of letters. They usedthis knowledge, apparently, to rationalize spelling–pronunciation pairs such asLK–elk, in which the letter name occurred in the initial position of the word. Letternames at the ends of words were less helpful, reflecting children’s greater attentionto the beginnings of words than to the ends. The results suggest that young childrenwho cannot recognize simple words can go beyond a logographic approach toreading when learning words that contain consonant letter name cues in salient po-sitions.

Bowman and Treiman’s (2002) finding that even children classified as pre-readers can sometimes go beyond a logographic method of linking print andspeech is surprising in light of the widespread view that such children initially ap-proach print in a nonalphabetic manner (e.g., Byrne, 1992; Ehri, 1998; Frith,1985). The goal of our study was to extend the findings of Bowman and Treimanand to provide a further test of the idea that prereaders are not limited to alogographic approach to reading and spelling. This study went beyond the work ofBowman and Treiman in several ways. First, we investigated letter name cluesconsisting of vowels rather than consonants. Specifically, we used two-letter spell-ings that corresponded to words with VC or CV pronunciations and that, in thephonetically motivated condition, had a vowel letter name at the beginning (e.g.,AP–ape) or at the end (e.g., PA–pay). In the arbitrary condition, the letters in thespellings did not correspond in a plausible way to the sounds in the spoken words.If children can use their knowledge of letter names in linking print and speech, atleast with letter names at the beginnings of words, then this ability should appearwith vowel letter names as well as consonant letter names. A second way in which

156 BOWMAN AND TREIMAN

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

our experiment went beyond previous work is that it included a spelling task and areading task that had very similar stimuli and procedures. This allowed for bettercomparisons between the two tasks than in previous research, which has relied oncross-experiment comparisons. Finally, our study examined children’s ability toremember the taught responses after a brief delay. This issue has been investigatedin some previous studies (e.g., Ehri & Wilce, 1985) but not others (e.g., Bowman &Treiman, 2002).

The participants in our study were preschoolers who could not read any of 22simple and highly frequent words such as look, stop, and in. Our reading screeningtest used the same words as that of Bowman and Treiman (2002), which were thewords that Ehri and Wilce (1985) had found to be easiest for beginners. Althoughwe refer to our participants as prereaders, following the lead of other researchers,the children may have been familiar with the spellings of their own first names orthe names of family members. However, these children were unable to decodeprinted words on the basis of spelling-to-sound correspondences—the hallmark ofalphabetic reading.

METHOD

Participants

Forty-two prereaders with a mean age of 4 years 9 months completed the experi-ment. Half of these children were randomly chosen to participate in the readingtask and half in the spelling task. Another 16 children began the experiment but didnot complete it. Six of these children were dropped from the study because theycould read at least one common word on the screening task, 9 (3 in the reading taskand 6 in the spelling task) found the task too demanding and at some point declinedto continue, and 1 was absent so often that he could not complete the sessions in atimely manner. Participants attended preschools in the Detroit, Michigan, metro-politan area that did not offer formal instruction in reading or writing. All of thechildren were native speakers of English.

Materials

There were four conditions: VC motivated, VC arbitrary, CV motivated, and CVarbitrary. The appendix lists the printed stimuli and the pronunciations for eachcondition. In the phonetically motivated conditions, both phonemes in the spokenwords were represented with plausible letters. The first vowel phoneme in the VCcondition and the final vowel phoneme in the CV condition were spelled with a let-ter that had the same spoken letter name. In the VC condition, for example, thespelling AP was paired with the word /ep/ (ape). The letters were reversed (PA) and

LOGOGRAPHIC READING AND SPELLING 157

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

paired with pay in the CV condition. To ensure that no child experienced duplicatespellings or pronunciations across conditions, we designed two sets of stimuli foreach condition. A given child was exposed to the VC stimuli from one set and theCV stimuli from the other. The vowels of the phonetically motivated stimuli wereA, E, I, and O. It was not possible to develop two sets of words with U that met therequirements of the study. For comparability with the Bowman and Treiman(2002) study, in which children learned five items per condition, each set also in-cluded a filler stimulus that had the letter Y in the target position. The fillers dif-fered in some respects from the other stimuli, so performance on the filler stimuliwas not analyzed.

The arbitrary condition paired the letters from one set of phonetically motivateditems with the pronunciations from the other set to produce two-letter stimuli witha vowel letter (or Y) in either the initial or final position. Within each set, the letterswere scrambled so that neither phoneme in the spoken word was spelled in a plau-sible manner. For example, ape was spelled OM in the arbitrary VC condition, andpay was spelled MO in the arbitrary CV condition. Children learned the arbitraryVCs from one set and the arbitrary CVs from the other set.

Most of the printed stimuli were nonwords, but the many constraints on thestimuli meant that it was necessary to include several words. The conventional pro-nunciation of these words was never taught as the targeted pronunciation for theexperimental stimuli. For example, IS appeared in the VC motivated condition, butit was pronounced as ice. If children knew the conventional pronunciations of thesewords, this may have made it harder for them to learn the pronunciations that weretaught in the experiment. Several steps were taken to guard against this possibility.Two of the most common real words in the set of stimuli (IT and IS) were includedon the reading screening test. As discussed earlier, we selected children who couldnot read any words on the screening list. We also pretested the children to deter-mine how they responded to the written and spoken stimuli before training began.The children never produced either the correct pronunciation or the target pronun-ciation for the real words included as stimuli, as mentioned below. Finally, theitems in the motivated condition were on average more frequent in texts designedfor kindergartners and first graders than the items in the arbitrary condition (Zeno,Ivens, Millard, & Duvvuri, 1995). Thus, any knowledge of the real words that maynot have surfaced during either the reading screening or the pretesting of the exper-imental stimuli should have made the motivated condition harder than the arbitrarycondition, the opposite of the predicted results.

The reading and spelling tasks used visually similar materials, facilitating com-parisons between the tasks. For spelling, children used 5-cm-high, yellow, foamrubber, uppercase letters. The 10 letters needed to spell the words in each set weredisplayed on a 38-cm-wide × 32-cm-high green and blue felt board. The board wasplaced horizontally in front of the child with the letters randomly laid out on theupper green portion and a 17 × 13 cm spelling board centered on the lower blue

158 BOWMAN AND TREIMAN

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

portion. The spelling board was made of plastic canvas covered in green felt andsurrounded by a red felt frame, leaving an opening in the center portion. A yellowvertical line divided the center portion into two side-by-side areas into which let-ters could be placed to spell the words. A small yellow arrow in the upper left cor-ner of the spelling board reminded the child where to place the first letter. Thecards for the reading task measured 17 × 13 cm and consisted of a green back-ground surrounded by a red frame with two 5-cm-high yellow block letters cen-tered in the open area.

To assess the children’s knowledge of letter names and sounds, we usedtwenty-six 10 × 15 cm cards, each with a single 5-cm-high uppercase letter in Arialfont. We also designed a letter name recognition task using thirteen 8 × 13 cmcards to assess children’s recognition of the 13 letters used in the experiment. Avertical line divided each card into two equal sections, each of which contained adifferent 4-cm-high letter printed in Arial font. Each of the letters appeared twice,once on the left half of a card and once on the right. Correct answers occurredequally often on the two sides.

Children were screened for reading ability using the same 22 simple words usedby Bowman and Treiman (2002). Two words and an easily identifiable colored pic-ture were displayed on each of eleven 14 × 22 cm cards, and the spatial arrange-ment of the words and the pictures differed across cards.

Procedure

Each child participated in the VC motivated, VC arbitrary, CV motivated, and CVarbitrary conditions of either the reading or spelling task over four sessions that av-eraged 2 to 3 days apart. A different condition was presented during each session,and the order of conditions and word sets was balanced across children. Each ses-sion began with the child choosing one of four puppets. The children were told thatthey would learn to read or spell the way the puppet did and that this was differentfrom how people read and spell. Next, children were pretested to evaluate how theyresponded to the written and spoken stimuli before training began. For the readingpretest, the cards were shown one at a time and the child was asked to read eachword. No feedback was given. For the spelling pretest, the puppet first showed thechild how to choose and place the letters. The puppet randomly selected a first let-ter, placed it in the opening by the yellow arrow, selected a second letter, andplaced it in the remaining opening. After showing the children how to select andplace the letters, the experimenter pronounced the first word, used it in a sentence,and then repeated it before encouraging the child to guess how to spell the puppet’sword. Once the child placed two letters, the puppet showed the child the correctspelling. This ensured that, as in the reading task, the children in the spelling tasksaw the correct spellings during the pretest. This process was repeated for each ofthe five words.

LOGOGRAPHIC READING AND SPELLING 159

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

After the child had been pretested on all five words for the session, the correctresponses were demonstrated. For the reading task demonstration, the puppetshowed each card, pointed to the word, said it, used it in a sentence, said it again,and asked the child to repeat the word. The children in the spelling task had alreadybeen exposed to the pronunciations of the words and the sentences. The spelling ofeach word was demonstrated, with the child repeating the words as they were dem-onstrated.

The demonstration phase was followed by up to eight learning trials. The read-ing task participants saw the cards presented one at a time in random order and had5 sec to respond to each. Correct answers were praised. If the child did not respondor gave an incorrect answer, the experimenter provided the correct answer. Thechildren in the spelling task heard each word pronounced and were encouraged toselect two letters to spell the word the way the puppet spelled it. Correct answerswere praised, and incorrect answers were replaced with the correct spelling.Learning trials continued until the child had achieved the criterion of respondingcorrectly to all five items on two consecutive trials (in which case we assumed thatthe child would have responded correctly on further trials) or until eight trials hadbeen completed.

A single screening or assessment task followed the learning trials of each ses-sion. These tasks occurred after the learning trials to provide a time delay afterwhich memory for the learned items was assessed once again. After the learningtrials of Session 1, the reading screening task was given. Children were shown theword and picture cards and were asked to identify anything familiar on the card. Ifthe child did not identify all three items, the experimenter pointed to each item andasked the child to identify it. Children were liberally praised for any correct re-sponses, including identification of the pictures, so that they would experiencesome success and be motivated to perform the task. The letter name and soundtasks were given after the learning trials of Sessions 2 and 3, their order balancedacross sessions. For the letter name task, the cards were presented one at a time in arandom order. Children were asked to name the letter on each card. If the correctresponse was not given in this free-choice situation, two letter names were sup-plied and the child was asked to pick the one that corresponded to the letter printedon the card. The letter sound task was similar except that children were asked forthe letters’ sounds. The letter name recognition task followed the learning trials ofSession 4. Children were shown each card, were told the name of a letter, and wereasked to point to the specified letter on the card. After the screening or assessmenttask in each session, which lasted approximately 5 min, children completed amemory task consisting of one final trial of reading or spelling the target words toassess how a brief delay affected memory for the learned items. At the end of thefourth session, children were shown the correct spellings of all the words used inthe study. They were again told that the puppets spelled the words differently fromhow people spell them.

160 BOWMAN AND TREIMAN

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

RESULTS

The pretest results showed that that the children never read and very rarelyspelled the words in the targeted manner before they were trained to do so. Thechildren also never produced the conventional pronunciations for any of thestimuli that were words. For example, AM was never pronounced with itstaught pronunciation, /em/ (aim), or its real pronunciation, /æm/ (am), beforethe learning trials began. Of interest then is how children learned the taught re-sponses across trials for the phonetically motivated and arbitrary stimuli andwhether they were able to remember these responses after a brief delay. Figures1 and 2 depict the proportion of taught responses across learning trials and inthe memory trial of the reading and spelling tasks, respectively. The data sug-gest that children showed better learning of the phonetically motivated pairsthan the arbitrary pairs and that a short delay did not markedly affect perfor-mance. The graphs further show that children showed more learning across tri-als in the reading task than the spelling task. The difference between the pho-netically motivated and arbitrary conditions was larger for the VCs, which hadthe vowel letter name at the beginning, than for the CVs, which had the lettername at the end.

LOGOGRAPHIC READING AND SPELLING 161

FIGURE 1 Proportion of correct responses on reading task as a function of condition, stimu-lus type, and trial. Note. V = vowel; C = consonant.

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

These impressions were confirmed by a repeated measures analysis of variancecarried out on the number of correct responses using the within-subject factors ofcondition (phonetically motivated vs. arbitrary), stimulus type (VC vs. CV), andlearning trial (1 to 8) and the between-subject factor of task (reading vs. spelling).We used a multivariate approach because the sphericity assumption was violated.This approach does not assume sphericity and is recommended by O’Brien andKaiser (1985). There was a main effect of condition, F(1, 40) = 44.54, p < .001,with children performing better in the phonetically motivated conditions than thearbitrary conditions. A main effect of trial was also observed, F(7, 34) = 15.64, p <.001, such that performance improved across the learning trials. These main effectswere qualified by two-way interactions between condition and stimulus type, F(1,40) = 10.73, p = .002, and condition and trial, F(7, 34) = 2.92, p = .017, as well as athree-way interaction involving condition, stimulus type, and trial, F(7, 34) = 3.70,p = .004. The difference between the phonetically motivated and arbitrary condi-tions increased across the learning trials. The three-way interaction reflected thefact that the difference between the phonetically motivated and the arbitrary condi-tions became larger for VCs, with the letter name at the beginning, than for CVs,with the letter name at the end.

162 BOWMAN AND TREIMAN

FIGURE 2 Proportion of correct responses on spelling task as a function of condition, stimu-lus type, and trial. Note. V = vowel; C = consonant.

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

Task participated in only one significant effect—an interaction of trial andtask, F(7, 34) = 2.36, p = .045. Children showed more improvement across thelearning trials in the reading task than the spelling task. Although the differencebetween the phonetically motivated and arbitrary conditions appeared to belarger for spelling than reading, the interaction between condition and task wasnot significant.

Figures 3 and 4 show the total number of correct responses in each condition ofthe reading and spelling tasks, respectively, along with the 95% confidence inter-val for the means. We adjusted for between-subjects differences in computing theconfidence intervals as recommended by Bakeman and McArthur (1996). The fig-ures show that phonetically motivated items were easier for children to learn thanarbitrary items, particularly for VCs. Further evidence of this comes from examin-ing the number of children who achieved the criterion of correctly responding to allfive items on two consecutive learning trials. Six children did so in the VC moti-vated reading condition and four in the CV motivated condition. The figures for thespelling condition were four and three, respectively. No child reached the criterionin the arbitrary condition of either the reading or spelling task. These data supportthe idea that the phonetically motivated items were easier for children to learn thanthe arbitrary items and that letter names in the initial position were more helpfulthan final letter names.

LOGOGRAPHIC READING AND SPELLING 163

FIGURE 3 Total number of correct responses (of 40 possible) in learning trials of readingtask as a function of condition and stimulus type; error bars represent 95% confidence intervals.Note. V = vowel; C = consonant.

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

Children showed the same pattern of performance on the final memory trials ason the learning trials, with better performance on phonetically motivated than arbi-trary items, F(1, 20) = 10.15, p = .005, for reading; F(1, 20) = 23.22, p < .001, forspelling. Paired t tests comparing performance on the memory trial and the finallearning trial revealed that, in each condition, performance after a 5-min delay wasstatistically equivalent to performance on the final learning trial. Thus, childrenshowed some memory for the responses they had learned after a short delay.

We scored children’s performance on the letter name task by both a strict and alenient system. According to the strict scoring system, children had to answer cor-rectly in the free-choice situation to receive credit. According to the lenient sys-tem, children received credit for a letter if they answered either the free-choice orthe two-choice question correctly. The mean proportion of correct responses on theletter name task was .72 by the strict scoring system and .91 by the lenient system.Performance was poorer in the letter sound task than the letter name task. Childrenaveraged .36 correct on the letter sound test by the strict scoring system and .73 bythe lenient system. The mean proportion correct on the forced choice letter namerecognition task was quite high (.95). These results show that the preschoolerstested here were rather knowledgeable about letter names and less knowledgeableabout letter sounds, the same pattern of results that has been found in other studieswith U.S. children (e.g., Worden & Boettcher, 1990).

164 BOWMAN AND TREIMAN

FIGURE 4 Total number of correct responses (of 40 possible) in learning trials of spellingtask as a function of condition and stimulus type; error bars represent 95% confidence intervals.Note. V = vowel; C = consonant.

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

If children use their knowledge of letter names and sounds to help learn to readand spell the phonetically motivated pairs, then measures of letter name and lettersound knowledge should correlate more highly with performance in the phoneti-cally motivated conditions than with performance in the arbitrary conditions. Thedata relevant to this prediction are shown in Table 1, which presents the correla-tions for all 42 children who completed the experiment. Because the patterns ofcorrelations were quite similar for the reading and spelling tasks, we pooled thedata across the tasks to maximize the sample size. The results shown in Table 1 arebased on a strict scoring of letter name and sound knowledge, as letter nameknowledge approached ceiling when scored by the lenient system. Children whoknew more letter names and sounds tended to perform better in all conditions ofthe word learning tasks than children who knew fewer letter names and sounds.However, as expected under the view that alphabetic knowledge is particularly im-portant for learning motivated associations, letter name and sound knowledgetended to correlate more highly with performance in the phonetically motivatedconditions than with performance in the arbitrary conditions. Specifically, lettersound knowledge correlated significantly more highly with performance in themotivated VC condition than in the arbitrary VC condition (p < .001, one-tailed)and significantly more highly with performance in the motivated CV conditionthan with performance in the arbitrary CV condition (p = .018, one-tailed). Lettername knowledge also correlated significantly more highly with performance in themotivated VC condition than with performance in the arbitrary VC condition (p =.037, one-tailed). A trend in the same direction was observed for the motivated andarbitrary CV conditions, but it was not statistically reliable.

DISCUSSION

According to many views of literacy development (e.g., Byrne, 1992; Ehri, 1998;Frith, 1985), young children initially approach the task of learning to read in alogographic way. Prereaders, according to these theories, are heedless of the linksbetween letters and sounds that are the foundation of alphabetic writing systems.

LOGOGRAPHIC READING AND SPELLING 165

TABLE 1Correlations Between Letter Name and Letter Sound Knowledge

(Strict Scoring) and Number of Correct Responses in Different Conditionsof Word Learning Task, Pooling Over Reading and Spelling Tasks

Measure Motivated VC Motivated CV Arbitrary VC Arbitrary CV

Letter names .53** .47** .30 .31*Letter sounds .73** .69** .39* .48**

*p < .05, two-tailed. **p < .01, two-tailed.

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

Our results, together with those of Bowman and Treiman (2002), raise questionsabout this view as applied to 4-year-old prereaders in the United States. These chil-dren, who typically know some letter names and sounds, use this knowledge tohelp learn simplified spellings such as AP for ape and MA for may and to remem-ber these spellings across at least a brief delay. Spellings that deviate from whatchildren know about letters, such as OM for ape and PO for may, are harder forthem to learn. If children relied solely on a logographic or prealphabetic approachto link print and speech, such differences should not have been found. We did notinclude spellings with visually distinctive letters in the present study, and so we donot know whether the children would have shown a benefit for these. Clearly,though, the children did benefit from mappings between printed and spoken wordsthat were consistent with the English system of letter names. It is possible that chil-dren who are younger or less knowledgeable about letters than those tested herewould do no better on phonetically motivated pairs than on arbitrary pairs. How-ever, any such logographic period must be briefer than often thought.

In our study, like that of Bowman and Treiman (2002), children showed a largerdifference between phonetically motivated and arbitrary print–sound pairs whenthe motivated pairs had a letter name clue at the beginning (e.g., AP for ape, LK forelk) than when the motivated pairs had a letter name clue at the end (e.g., MA formay, FL for fell). This pattern thus applies to consonants, as tested in the earlierstudy, and vowels, as tested here. Children’s greater attention to the beginningsthan the ends of words demonstrates another type of knowledge that is presenteven among prereaders: knowledge of the left-to-right directionality of Englishwriting. Prereaders pay more attention to the beginning than the end letters of theirown names (Treiman, Cohen, Mulqueeny, Kessler, & Schechtman, 2007), and theresults presented here suggest that the same is true for other words.

One difference between these results and those of Bowman and Treiman (2002)is that the superiority for the phonetically motivated condition over the arbitrarycondition was not statistically significant for final letter names in the previousstudy with consonants but was significant with the vowels tested here. This differ-ence may arise because the letter names used in our study are more familiar toyoung children than the letter names used in the earlier study. The mean proportionof correct responses for the vowel letter names used here was .70 in a large study ofU.S. preschoolers (Treiman, Kessler, & Pollo, 2006), as compared to .62 for the L,N, and R used by Bowman and Treiman. The more familiar children are with thename of a letter, the more readily they may use this knowledge in linking printedand spoken words. That is, the vowel letters tested here may have yielded differentresults than the consonant letters tested previously because of differences in famil-iarity, not because of intrinsic differences between vowels and consonants per se.

The prereaders in our study performed better on phonetically motivated pairs thanarbitrary pairs in both reading and spelling tasks. We saw a trend toward more benefitfrom alphabetically motivated associations in spelling than in reading, consistent with

166 BOWMAN AND TREIMAN

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

Frith’s (1985) view that children use motivated associations in spelling before theydo in reading. However, the trend was not statistically reliable. The major differ-ence between the reading and spelling tasks was that spelling was harder. Previouscomparisons between reading and spelling tasks, although hindered by a lack ofcomparability of the stimuli and because the tasks were administered in separateexperiments, have also found poorer performance in the sound-to-spelling direc-tion (Bowman & Treiman, 2002; Treiman et al., 2001). Indeed, spelling continuesto be more difficult than reading for older children than adults. The fact that thechildren in our study performed better on phonetically motivated pairs than arbi-trary pairs in both spelling and reading tasks supports the generality of our conclu-sions. Children who have very limited reading skills but who are familiar with let-ters need not be limited to a logographic approach in learning about print.

The idea that familiarity with letters helps children go beyond a logographic ap-proach to reading and spelling is supported by our correlational results. Knowl-edge of letter names and sounds generally correlated more highly with ability tolearn phonetically motivated spelling–sound pairs than with ability to learn arbi-trary pairs. Treiman et al. (2001) and Bowman and Treiman (2002) reported simi-lar patterns of correlations. These findings suggest that, as young children gain fa-miliarity with the names and sounds of letters, their ability to benefit fromalphabetically motivated links between spellings and sounds improves. These im-provements occur even before children gain the ability to recognize commonwords such as go and it or to decode previously unseen words. As Rack et al.(1994) suggested, the better learning of phonetically motivated than arbitrary pairsis consistent with connectionist models of word reading. These models, as re-viewed by Plaut (2005), learn mappings between orthographic and phonologicalunits through repeated exposure to the conventional pronunciations of printedwords. In future work, it will be important to compare the course of learning inchildren with the course of learning in specific connectionist models, using thesame stimuli.

Examination of reading and spelling performance over the course of training isvaluable not only for the testing of models but also for practical reasons. Spe-cifically, looking at a child’s learning over a series of trials may provide a better in-dication of the child’s potential than looking at performance on a single trial, as re-searchers and educators typically do. When presented with an isolated word andasked to read or spell it, children of the age of those in this study perform quitepoorly. The children tested here rarely read or spelled even the phonetically moti-vated words on the pretest, and they never correctly identified printed words suchas eat and go on our reading screening test, even though these words include letternames. The results of our training study suggest that children could have learnedthe pronunciations of such letter name words with extensive practice, and moreeasily than they could have learned the pronunciations of words such as is and who.But children of this age learn even phonetically motivated words quite slowly, and

LOGOGRAPHIC READING AND SPELLING 167

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

preschoolers do not usually get the intensive training that would be necessary forthem to learn and remember words. Examining how a child responds to such train-ing may provide an indicator of the child’s potential for success in early literacy in-struction, as well as shedding light on the skills the child needs to learn in order toachieve success.

In our study, we used a rote, paired-associate training procedure to ensure com-parability across conditions. We do not claim, however, that this is the best way toteach young children to read and spell words. Had the connections between theprinted letters and the spoken sounds in the phonetically motivated spellings beenhighlighted through teaching, the superiority for the motivated condition over thearbitrary condition might have been larger than observed here. Our results suggestthat a number of preschoolers have knowledge that could help them benefit fromsuch teaching.

Several researchers have raised questions about the importance and duration ofa logographic stage of literacy development for children who are exposed to lan-guages in which the links between spellings and sounds are easier to notice or moresystematic than they are in English. For example, Cardoso-Martins et al. (2002)raised such questions for Portuguese, and Wimmer and Hummer (1990) did thesame for German. Our results suggest that the importance and duration of alogographic stage have been overestimated, as well, for children who are exposedto English. Our findings suggest that U.S. 4-year-olds who cannot read any simplewords have more reading-related skills than is often acknowledged. Although acomplete understanding of the alphabetic system is certainly some distance awayfor these young children, prereaders are not always limited to making rote,logographic associations between letter strings and pronunciations.

ACKNOWLEDGMENTS

This research was partially supported in part by National Science FoundationGrant BCS-0130763 and National Institutes of Health Grant HD51610.

REFERENCES

Abreu, M. D. de, & Cardoso-Martins, C. (1998). Alphabetic access route in beginning reading acquisi-tion in Portuguese: The role of letter-name knowledge. Reading and Writing, 10, 85–104.

Bakeman, R., & McArthur, D. (1996). Picturing repeated measures: Comments on Loftus, Morrison,and others. Behavior Research Methods, Instruments, & Computers, 28, 584–589.

Bowman, M., & Treiman, R. (2002). Relating print and speech: The effects of letter names and wordposition on reading and spelling performance. Journal of Experimental Child Psychology, 82,305–340.

168 BOWMAN AND TREIMAN

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

Byrne, B. (1992). Studies in the acquisition procedure for reading: Rationale, hypotheses, and data. InP. B. Gough, L. C. Ehri, & R. Treiman (Eds.), Reading acquisition (pp. 1–34). Hillsdale, NJ: Law-rence Erlbaum Associates.

Cardoso-Martins, C., Resende, S. M., & Rodrigues, L. A. (2002). Letter name knowledge and the abil-ity to learn to read by processing letter-phoneme relations in words: Evidence from Brazilian Portu-guese-speaking children. Reading and Writing, 15, 409–432.

Ehri, L. C. (1998). Grapheme–phoneme knowledge is essential for learning to read words in English. InJ. L. Metsala & L. C. Ehri (Eds.), Word recognition in beginning literacy (pp. 3–40). Mahwah, NJ:Lawrence Erlbaum Associates.

Ehri, L. C., & Wilce, L. S. (1985). Movement into reading: Is the first stage of printed word learning vi-sual or phonetic? Reading Research Quarterly, 20, 163–179.

Foulin, J. N. (2005). Why is letter-name knowledge such a good predictor of learning to read? Readingand Writing, 18, 129–155.

Frith, U. (1985). Beneath the surface of developmental dyslexia. In K. E. Patterson, J. C. Marshall, &M. Coltheart (Eds.), Surface dyslexia: Neuropsychological and cognitive studies of phonologicalreading (pp. 301–330). London: Lawrence Erlbaum Associates.

International Phonetic Association. (1999). Handbook of the International Phonetic Association: Aguide to the use of the International Phonetic Alphabet. Cambridge, England: Cambridge UniversityPress.

Laing, E., & Hulme, C. (1999). Phonological and semantic processes influence beginning readers’abil-ity to learn to read words. Journal of Experimental Child Psychology, 73, 183–207.

O’Brien, R. G., & Kaiser, M. K. (1985). MANOVA method for analyzing repeated measures designs:An extensive primer. Psychological Bulletin, 97, 316–333.

Plaut, D. C. (2005). Connectionist approaches to reading. In M. J. Snowling & C. Hulme (Eds.), Sci-ence of reading: A handbook (pp. 24–38). Oxford, England: Blackwell.

Rack, J., Hulme, C., Snowling, M., & Wightman, J. (1994). The role of phonology in young childrenlearning to read words: The direct-mapping hypothesis. Journal of Experimental Child Psychology,57, 42–71.

Treiman, R., Cohen, J., Mulqueeny, K., Kessler, B., & Schechtman, S. (2007). Young children’s knowl-edge about printed names. Child Development, 78, 1458–1471.

Treiman, R., Kessler, B., & Pollo, T. C. (2006). Learning about the letter name subset of the vocabulary:Evidence from U.S. and Brazilian preschoolers. Applied Psycholinguistics, 27, 211–227.

Treiman, R., & Rodriguez, K. (1999). Young children use letter names in learning to read words. Psy-chological Science, 10, 334–338.

Treiman, R., Sotak, L., & Bowman, M. (2001). The roles of letter names and letter sounds in connectingprint and speech. Memory & Cognition, 29, 860–873.

Wimmer, H., & Hummer, P. (1990). How German speaking first graders read and spell: Doubts on theimportance of the logographic stage. Applied Psycholinguistics, 11, 349–368.

Worden, P. E., & Boettcher, W. (1990). Young children’s acquisition of alphabet knowledge. Journal ofReading Behavior, 22, 277–295.

Zeno, S. M., Ivenz, S. H., Millard, R. T., & Duvvuri, R. (1995). The educator’s word frequency guide.Brewster, NY: Touchstone Applied Science Associates.

LOGOGRAPHIC READING AND SPELLING 169

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011

APPENDIX

Printed Stimuli and Pronunciations for Learning Tasks,With Filler Pairs Listed in Parentheses

VC phonetically motivated conditionSet 1: AP ape, EL eel, IS ice, OT oat, (YD wide)Set 2: AM aim, EK eek, IL aisle, OD owed, (YF wife)

VC arbitrary conditionSet 1: OP aim, YL eek, AS aisle, IT owed, (ED wife)Set 2: OM ape, YK eel, AL ice, ID oat, (EF wide)

CV phonetically motivated conditionSet 1: MA may, KE key, LI lie, DO doe, (FY fry)Set 2: PA pay, LE Lee, SI sigh, TO toe, (DY dry)

CV arbitrary conditionSet 1: MO pay, KY Lee, LA sigh, DI toe, (FE dry)Set 2: PO may, LY key, SA lie, TI doe, (DE fry)

170 BOWMAN AND TREIMAN

Dow

nloa

ded

by [

Reb

ecca

Tre

iman

] at

07:

53 0

8 Ju

ly 2

011